Minutes, May 18, 2000 OHS/DKR Meeting at SRI, taken by Howard Liu
In attendance: Su-Ling Yee, Mary Coppernoll, Doug Engelbart, Warren Stringer, Pat Lincoln, Eric Armstrong, Jack Park, Mary Keeler, John Deneen, Rod Welch, Sonny Kirkley, Joe Williams, Howard Liu, Lee Iverson, Cynthia Waddell
Welcomes
Welcome Sonny Kirkley, information in place, inc., Indiana
Welcome invited speaker Mary Keeler, researcher at University of Washington
News
Doug, Pat, Lee, Eric, Eugene met with SourceForge (i.e. VA Linux), portal website for open source software development
Lee: Only requirement for project to be open source (i.e. stamped by OSI)
Want to develop hosting services to large companies that seek hosting service within company
Interested in better tools for software development
Keen on allocation of resources
Can host our development project, or at least track its progress
Doug or Lee invited to upcoming seminar
Doug
Possible to invite our whole group?
Lee
Will investigate
Lee
Other news?
Rob
What knowledge means. Morris and Eric and Rob discussed knowledge management at Intel
Impetus: Bellinger working definitions to guide development
Doug
Need glossary
Joe
Difference between transcode and translate
Lee
Zope is installed on bootstrap.org
Will announce when available to users
John Deneen
Gary Baldwin, tech director, GigaSystems
Berkeley Wireless System
John will bring in their licensing agreement for reference
Doug
Effort to connect Academia Sinica
Doug
Adobe
IBM Almaden
Rob
0.6 requirements exist
Lee
Must think through requirements more carefully
Eric
Narrative is fundamental to object-oriented design
Find nouns and verbs in narrative
Lee
Eugene is working on use cases
Higher level architecture based on Jack and Howard’s meeting with Doug
Jack
Developing architecture is evolving process
Agenda: invited speaker Mary Keeler
Mary: who is the audience?
Lee: SRI, interested people attracted by desire to use the OHS
Mary: How does OHS relate to the DKR?
Lee: Need means to enable and track collaboration for document preparation, management
Mary: How does OHS relate to W3C?
Lee: we want to build on what’s available from W3C
Lee: John Bozak of W3C is interested
Welcome Cynthia Waddell, San Jose City Manager Dept, law and policy, paper on Disability, compliance officer for American Disability’s Act
Doug
Digital government
GSA
Doug
Neil Scott, Stanford, is working on user interfaces for disabled people
Mary Keeler speaks (cont’d)
Lee
No need to limit to documents in traditional sense
Mary
see whiteboard
Jack
setting context for her discussion to apply
Doug: What’s an ontology
Mary: an area in classical philosophy that has been abandoned by modern philosophy
The study of what really is there
Existence and reality
Being and existence
Mary
Ancients asked what exists
Moderns gave up questions on existence
Modern philosophers deal only with semantic relations with represented entities---modern logic
KR people picked up terms, e.g. "entity," "attribute," from ontology
Scientist leap to expressed evidence, and take forms for existence
Should go back to find more evidence to be responsible for the existence level
Mary
Raw data; types; symbols
Sequence, not hierarchy: tone, types, token
Possibility, actuality, probability
Potentiality, types, symbols
Blur, lens, image
Raw data: we consider these as the foundation in the DKR, rather than a yet lower level
But we don’t limit documents to mere text
Want to get to the symbols
OHS gets at the symbols
Flow of information
Arguments expressed by OHS users are raw data, so will sink to raw data layer
Will bubble up to symbols again
Logic
Peirce
Lens distorts and help
Role of logic: Logic is a lens
Aristotle
Jack: note, compare Peirce’s notion that theories are always improving parallels Doug’s notion of the OHS/DKR’s evolutionary development
Categories and relations
Because of lens, can see things in blur, but also miss things
Must go back and forget lens focus to see what’s there
"The more precise, the less you see."
Warren: psychology report confirms that for human perception
Think of stages of improving the imperfect lens
In intellectual development, we create new lenses also
Traditional logic does nothing about existence
Peirce: only when we built mechanism to do reality check do we get reliability; just checking validity for relationship between forms is not enough.
Validity is soundness for relations of symbols
Reliability is soundness for relations of symbols grounded in raw data, or what exists
So therefore Conceptual Graph
An example of a means to create lens
A graphical notation to express logic
A lens in itself
Other means exist
Mary (aside):
Peirce liked triads
The notion of Generation is expressed; one cannot express the generation notion without three: two won’t do it
Affector, Affected, and an abstract relation---the act of affecting
Note, the abstract relation is artificial, abstract, and not on the same order of things as Affector and Affected.
How DKR can be that mechanism to do reality check for reliability
The types layer is the lens
Raw data (tone) is messy and left out by positivist, but we must consider that layer
Possibility, actuality, probability
Raw data, e.g. sounds in the room, blur, undifferentiated
In top layer: has committed bias, user world view
In middle layer: tries to be neutral, though never can be, but keep trying to be better at been neutral
Just as a lens can never be perfect
The lens maker keeps trying to evolve a better, more neutral lens
How does phenomenology fit in?
some philosophers wanted to replace Aristotle’s metaphysics with phenomenology
appearance: if we don’t know what things really are, we can deal only with how things appear---positivist
We in the DKR take documents as the existence layer
Lee: people not deal with what is, but what is reported
There’s no perfect lens
we must keep improving the lens
we must keep developing the lens knowing imperfect and will keep improving
we use the lens with the knowledge of its imperfection
Raw data: anything can relate to anything else---we don’t want to establish relations at the raw data layer, i.e. the layer of possibilities
Actuality middle layer: establish relations
Doug: can conceptual graph…?
Mary: Existential graph lots richer than Conceptual Graphs (CG)
Middle layer: impose lens on the blur of potentiality (raw) bottom layer
Middle layer: holds the Conceptual Graphs, which translate readily to natural language, impose a knowledge representation (KR)
Raw data, types, rules---more familiar terms to KR community
Raw: undifferentiated experience---raw data
Caution: communication should not be expected to, and need not, result in agreement
It isn’t a goal to bring different views to the same view
Goal: Must enable clear expression of all those (possibly) different views that they can be compared and talked about
Collaboration should not just come to agreement,
Never should sacrifice diversity to agreement
Instead collaboration should facilitate growing the thing to something greater than itself
Cyclic operation: imperfect lens always, must keep checking for its imperfections so to improve it
Scientists should keep refreshing what they see
Eric: Requirements, design, implementation in software development
Alternatives
If implementation fails, then user can go back to requirements
Lee: collaboration means bringing voices together, tool should expose where consensus exists and where disagreement exists. Tool guarantees validity, not reliability
Want the system to do the boring logic, so human can do more interesting things
Conceptual graph----Alabama. Look for cg community there
Cg systems exist
Cg just an instrument, just an interface that makes apparent
Cg good for machine and human to understand
Think of knowledge as structurally richer data